04. Data Science and Stochastic Optimization
The Armed Forces (FFAA) have a considerable mass of financial and biometric data of its pensioners that needs to be submitted to its own actuarial mathematical model, in order to obtain the result of the actuarial projection of 75 years, which is a legal obligation that must be met annually. The aim of the study is to reduce the computational times necessary to obtain the results of this projection, through the application of Parallel Computing and Software Engineering techniques. It is known that actuarial calculus is a mathematical method that uses financial, economic and probabilistic concepts to size the amount of resources and contributions necessary to pay future benefits of policyholders. The research presented here achieved significant results, in that it reduced by more than 90% the processing times of the calculations referring to the actuarial projections of FFAA pensioners. In the specific case of the Brazilian Army, from a mass of data of 500,000 records, it was possible to execute a complex mathematical model of high recursivity in about 5 minutes. Currently, the world goes through numerous transformations, generating new demands for society. Problems grow in quantity and complexity, in an increasingly dynamic and volatile context, requiring quick decisions from managers, especially at the strategic level. Hence the importance of the speed of actuarial calculus under study, because, thus, it can be timely trace dearly in order to support the decision of the FFAA Senior Management.
PALAVRAS-CHAVE: parallel computing, time reduction, actuarial calculus, armed forces, actuarial projection.